intel server processors: performance-per-watt · pdf fileintel server processors:...

INTEL SERVER PROCESSORS: PERFORMANCE-PER-WATT COMPARISON

MARCH 2010

A PRINCIPLED TECHNOLOGIES TEST REPORT Commissioned by Intel Corp.

OUR FINDINGS As data centers reach their power and cooling

capacities, efficiency has become a key focus. The

new Intel® Xeon® processor 5600 series delivers

better performance per watt and higher

performance than its predecessors. In Principled

Technologies’ tests in our labs, a two-socket server

with the new Intel® Xeon® Processor X5670

demonstrated significant advantages over the

previous-generation Intel Xeon Processor X5570-

based server in both these areas.

OUR PROCESS

We used the industry-standard SPEC® CPU2006

benchmark to focus on and measure the processor

performance of a pair of servers. While that

benchmark was running, we captured the energy

usage of each server. With this data, we computed

both relative performance and performance per

watt.

A Principled Technologies Test Report 2

Intel server processors: Performance-per-watt comparison

WORKLOAD The SPEC CPU2006 workload includes two benchmark suites: CINT2006 and CFP2006. (Note: SPEC and

SPECint are trademarks of the Standard Performance Evaluation Corporation.) We ran only the CINT2006

benchmark, which focuses on measuring and comparing compute-intensive integer performance. Specifically,

we measured the SPECint_rate_base2006 results for the test servers with 16 and 24 users.

Generally, the best SPECint_rate_base2006 score occurs using the same number of users as execution

units for a given server. The optimum user count for our testing was 24 users on the Intel Xeon Processor

X5670-based server and 16 users on the Intel Xeon Processor X5570-based server. The difference in user

counts between the servers is due to the different number of execution units (logical or physical processors)

on those servers.

Figure 1 lists the 12 applications that compose the CINT2006 benchmark. SPEC wrote nine of the

applications in C and three (471.omnetpp, 473.astar, 483.xalancbmk) in C++.

A CINT2006 run performs each of the 12 applications three times and reports the median for each. It

also calculates the geometric mean of those 12 results to produce an overall score. (For more information on

SPEC CPU2006 and other SPEC benchmarks, see www.spec.org.)

Name Application area

400.perlbench Programming language

403.gcc C compiler

429.mcf Combinatorial optimization

445.gobmk Artificial intelligence: Go

456.hmmer Search gene sequence

458.sjeng Artificial intelligence: chess

462.libquantum Physics/quantum computing

464.h264ref Video compression

471.omnetpp Discrete event simulation

473.astar Path-finding algorithms

483.xalancbmk XML processing

Figure 1: The applications that make up the CINT2006 benchmark.



SYSTEM COMPARISON Figure 2 shows a side-by-side comparison of the key hardware differences between the two systems.

Appendix A provides detailed configuration information.

Hardware specifications Intel Xeon Processor X5570-

based system Intel Xeon Processor X5670-

based system

CPU Intel Xeon Processor X5570 Intel Xeon Processor X5670

CPU speed (GHz) 2.93 GHz 2.93 GHz

Number of processor packages 2 2

Number of cores per processor package

4 6

Number of hardware threads per core 2 2

Memory type PC3-10600R PC3-10600R

Total memory (GB) 24 GB (6 x 4 GB) 24 GB (6 x 4 GB)

Hard drive 1 x Western Digital WD160 1 x Western Digital WD160

Figure 2: System configuration information for the two systems.

WHAT WE FOUND

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

No

rmalized

co

mp

ari

so

n

Server

SPECint_rate_base2006 relative performance/watt results

Intel Xeon X5670-based server


Figure 3: Normalized SPECint_rate_base2006 performance/watt results for the two servers. Higher numbers are better.

As Figure 3 shows, the Intel

Xeon Processor X5670-based server

delivers 28.9 percent more

performance per watt than the Intel

Xeon Processor X5570-based server.

To calculate the

performance/watt, we used the

following formula: benchmark score

divided by average peak power

consumption in watts. We normalized

the results to the Intel Xeon X5570-

based server.



Figure 6 details the results of our tests with the optimum number of users for SPECint_rate_base2006.

We determined the number of users based on the number of execution units in a given server. We used the

same number of SPECint_rate_base2006 users as processor execution units, so there is a one-to-one ratio.

SPECint_rate_base2006 performs three runs of each benchmark in the test suite and records the median, so

the final score is a median of three runs. Higher scores are better.

0

50

100

150

200

250

300

350

Sco

re

Server

SPECint_base_rate2006 results





achieved a 35.2 percent greater

SPECint_rate_base2006 score than the

Intel Xeon Processor X5570-based

server, with respective scores of 319

and 236.

Figure 4: SPECint_rate_base2006 results for the two servers. Higher numbers are better.

0

50

100

150

200

250

300

350

400

Watt

s

Server

SPECint_rate_base2006 power results





used only 4.8 percent more power

than the Intel Xeon Processor X5570-

based server while running the

SPECint_rate_base2006 benchmark,

with respective scores of 368.6 watts

and 351.6 watts.

Figure 5: SPECint_rate_base2006 power results for the two servers. Lower numbers are better.



Figure 6 also details the power consumption, in watts, of the test servers while idle and during the

benchmark. Idle power is an average of a 2-minute power recording while the server was idle. Peak power

represents the average power for the duration of the benchmark run.

Server SPECint_rate_base

2006 results Idle

power Peak

power User

count

Intel Xeon Processor X5570-based server

236 153.8 351.6 16


319 157.7 368.6 24

Figure 6: SPECint_rate_base2006 results; power consumption, in watts; and user count of the test servers while idle and during the benchmark.

HOW WE TESTED

Intel configured and provided the two Intel Xeon processor-based servers.

We began by installing a fresh copy of SUSE™ Linux Enterprise Server 11 Service Pack 2. We installed

the default packages, the C/C++ Compilers and Tools, and disabled the firewall. To maximize performance per

watt for each server, we adjusted the BIOS settings by disabling Turbo mode on both systems. We made no

additional changes to the default installation options.

SPECCPU2006 configuration

Intel compiled and provided the SPEC CINT2006 executables, but followed SPEC’s standard instructions

for building the executables using the following software tools for both servers:

Intel C++ Professional Compiler for IA32 and Intel 64, Version 11.1

MicroQuill SmartHeap V8.1

Binutils 2.18.50.0.7.20080502

The benchmark requires a configuration file. Intel provided the configuration file we used for the Intel

Xeon Processor X5570-based server and the Intel Xeon Processor X5670-based server. The configuration file

we used appears in Appendix B.

We report only the base metrics for the SPECint_rate test. SPEC requires the base metrics for all

reported results and sets compilation guidelines that testers must follow in building the executables for such

tests.



To begin the benchmark, we executed the following script on the Intel Xeon Processor X5570-based

server:

. ./shrc

ulimit -s unlimited

export LD_LIBRARY_PATH=$SPEC/libic11.1-32bit/:$SPEC/libic11.1-

64bit/:$LD_LIBRARY_PATH

a=`cat /proc/cpuinfo | grep processor | wc -l`

rm -rf topo.txt

specperl nhmtopology.pl

b=`cat topo.txt`

echo "*****************************************"

echo Running rate with $a copies on a NHM system with a topology of $b

echo "*****************************************"

rm -rf benchspec/CPU2006/*/run

runspec --rate $a -c cpu2006.1.1.ic11.1.linux64.sse42.rate.jan182010.cfg --

flagsurl=Intel-ic11.1-linux64-revE.xml --define dp-nhm --define smt-on --

define $b -T base -o all int

We used the same script for the Intel Xeon Processor X5670, but changed the --define dp-nhm to -

-define dp-wsm. These commands tell the benchmark how many cores and define how it runs.

When the run completes, the benchmark puts the results in the directory /cpu2006/result. The result

file names are of the form CINT2006.<number>.<suffix>. The suffixes are html, asc, raw, and pdf. The number

is three digits and associates a result file with its log, e.g., CINT2006.002. asc and log.002.

Appendix C provides the SPECint_rate_base2006 output results for each of the two test servers.

Power measurement procedure

To record each server’s power consumption during each test, we used an Extech Instruments

(www.extech.com) 380803 Power Analyzer/Datalogger. We connected the power cord from the server under

test to the Power Analyzer’s output load power outlet. We then plugged the power cord from the Power

Analyzer’s input voltage connection into a power outlet.

We used the Power Analyzer’s Data Acquisition Software (version 2.11) to capture all recordings. We

installed the software on a separate Intel processor-based PC, which we connected to the Power Analyzer via

an RS-232 cable. We captured power consumption at 1-second intervals.

http://www.extech.com/



To gauge the idle power usage, we recorded the power usage for 2 minutes while each server was

running the operating system but otherwise idle.

We then recorded the power usage (in watts) for each server during the testing at 1-second intervals.

To compute the average power usage, we averaged the power usage during the entire SPECint_rate_base2006

run. We call this time the power measurement interval. See Figure 6 for the results of these measurements.



APPENDIX A – SERVER CONFIGURATION INFORMATION

Figure 7 provides detailed configuration information about the test systems.

Servers Intel Xeon Processor X5570-

based server Intel Xeon Processor X5670-

based server

Power supplies

Total number 2 2

Wattage of each (W) 885 885

Cooling fans

Total number 3 3

Dimensions (h x w) of each 3” x 3” 3” x 3”

General processor setup

Number of processor packages 2 2

Number of cores per processor package

4 6

Number of hardware threads per core

2 2

CPU

Vendor Intel Intel

Name Xeon X5570 Xeon X5670

Stepping D0 C0

Socket type LGA1366 LGA1366

Core frequency (GHz) 2.93 2.93

Bus frequency 6.4 GT/s 6.4 GT/s

L1 cache (KB) 32 + 32 (per core) 32 + 32 (per core)

L2 cache (KB) 256 KB (per core) 256 KB (per core)

L3 cache (MB) 8 12

Thermal design power (TDP, in watts)

95 95

Platform

Vendor and model number Supermicro SuperServer 6026T-NTR+

Supermicro SuperServer 6026T-NTR+

Motherboard model number Super X8DTN+ Super X8DTN+

Motherboard chipset Intel 5520 Intel 5520

BIOS name and version American Megatrends 4.6.3.2 (01/06/2010)

American Megatrends 4.6.3.2 (01/06/2010)

BIOS settings All default settings except Turbo mode disabled

All default settings except Turbo mode disabled

Memory modules

Total RAM in system (GB) 24 24



Servers Intel Xeon Processor X5570-

based server Intel Xeon Processor X5670-

based server

Number of types of memory modules

1 1

Speed in the system currently running @ (MHz)

1,333 1,333

Timing/Latency (tCL-tRCD-iRP-tRASmin)

9-9-9-24 9-9-9-24

Vendor and model number MT36JSZF51272PY MT36JSZF51272PY

Type PC3-10600R PC3-10600R

Speed (MHz) 1,333 1,333

Speed in the system currently running @ (MHz)

1,333 1,333

Timing/Latency (tCL-tRCD-iRP-tRASmin)

9-9-9-24 9-9-9-24

Size (GB) 4 4

Number of RAM modules 6 x 4GB 6 x 4GB

Chip organization Double-sided Double-sided

Hard disk

Vendor and model number Western Digital WD1600AAJS-00M0A0

Western Digital WD1600AAJS-00M0A0

Number of disks in system 1 1

Size (GB) 160 160

Buffer size (MB) 8 8

RPM 7,200 7,200

Type SATA SATA

Controller Intel Corporation ICH10R SATA 3.0Gbps Controller

Intel Corporation ICH10R SATA 3.0Gbps Controller

Operating system

Name SUSE Linux Enterprise Server 11 SUSE Linux Enterprise Server 11

File system ext3 ext3

Kernel 2.6.27.19-5 2.6.27.19-5

Language English English

Network card/subsystem

Vendor and model number Intel 82576EB Gigabit Ethernet Intel 82576EB Gigabit Ethernet

Type Integrated Integrated

USB ports

Number 4 4

Type USB 2.0 USB 2.0 Figure 7: Detailed configuration information about the test systems.



APPENDIX B – SPECint_rate_base2006 CONFIGURATION INFORMATION

# Invocation command line:

# /usr/cpu2006/bin/runspec --rate 16 -c

cpu2006.1.1.ic11.1.linux64.sse42.rate.jan182010.cfg --flagsurl=Intel-ic11.1-linux64-

revE.xml --define dp-nhm --define smt-on --define physicalfirst -T base -o all int

# output_root was not used for this run

############################################################################

############################################################################

# This is a sample config file. It was tested with:

#

# Compiler name/version: Intel Compiler 11.1

# Operating system version: 64-Bit SUSE LINUX Enterprise Server 10 or later

# Hardware: Intel processors supporting SSE4.2

#

############################################################################

# SPEC CPU2006 Intel Linux64 config file

# Sep 2009 IC 11.1 Linux64

############################################################################

action = validate

tune = base

ext = cpu2006.1.1.ic11.1.linux64.sse42.rate.jan182010

PATHSEP = /

check_md5=1

reportable=1

bench_post_setup=sync

#

# These are listed as benchmark-tuning-extension-machine

#

int=default=default=default:

CC= icc -m32

CXX= icpc -m32

OBJ = .o

SMARTHEAP32_DIR = /home/cmplr/usr3/alrahate/cpu2006.1.1.ic11.1/libic11.1-32bit

SMARTHEAP64_DIR = /home/cmplr/usr3/alrahate/cpu2006.1.1.ic11.1/libic11.1-64bit

fp=default=default=default:

CC= icc -m64

CXX= icpc -m64

FC= ifort -m64

OBJ = .o

# For UP systems, we need to know if the processors are ordered across cores first or in

order

# If across cores, processors 0, 1, 2 and 3 are on distinct physical cores

# Otherwise, processors 0, 2, 4 and 6 are on distinct physical cores

default:

submit = numactl --localalloc --physcpubind=$SPECCOPYNUM $command

%ifdef %{no-numa)

submit = taskset -c $SPECCOPYNUM $command

%endif

####################################################################



# Compiler options

# for Nehalem use -xSSE4.2

# for processors prior to dunnington, replace -xSSE4.1 with -xSSSE3

####################################################################

default:

SSE = -xSSE4.2

FAST = $(SSE) -ipo -O3 -no-prec-div -static

FASTNOSTATIC = $(SSE) -ipo -O3 -no-prec-div

################################################################

#

# portability & libraries

#

#################### Portability Flags and Notes ############################

400.perlbench=default:

CPORTABILITY= -DSPEC_CPU_LINUX_IA32

403.gcc=default:

EXTRA_CFLAGS= -Dalloca=_alloca

462.libquantum=default:

CPORTABILITY= -DSPEC_CPU_LINUX

483.xalancbmk=default:

CXXPORTABILITY= -DSPEC_CPU_LINUX

fp=default:

PORTABILITY = -DSPEC_CPU_LP64

435.gromacs=default=default=default:

LDPORTABILITY = -nofor_main

436.cactusADM=default=default=default:


454.calculix=default=default=default:


481.wrf=default=default=default:

CPORTABILITY = -DSPEC_CPU_CASE_FLAG -DSPEC_CPU_LINUX

################################################################

# Tuning Flags

################################################################

#

# Base tuning default optimization

# Feedback directed optimization not allowed in baseline for CPU2006

# However there is no limit on the number of flags as long as the same

# flags are used in the same order for all benchmarks of a given language

471.omnetpp,473.astar,483.xalancbmk=default:

EXTRA_LIBS= -L$(SMARTHEAP32_DIR) -lsmartheap

EXTRA_LDFLAGS= -Wl,-z,muldefs

int=base=default=default:



COPTIMIZE= $(FAST) -opt-prefetch

CXXOPTIMIZE= $(FASTNOSTATIC) -opt-prefetch

fp=base=default=default:

OPTIMIZE= $(FAST)

################################################################

# Peak Tuning Flags int 2006 fast

################################################################

int=peak=default:

COPTIMIZE= -auto-ilp32 -ansi-alias

CXXOPTIMIZE= -ansi-alias

PASS1_CFLAGS = -prof-gen

PASS2_CFLAGS = $(FAST) -prof-use

PASS1_CXXFLAGS = -prof-gen

PASS2_CXXFLAGS = $(FASTNOSTATIC) -prof-use

PASS1_LDCFLAGS = -prof-gen

PASS2_LDCFLAGS = $(FAST) -prof-use

PASS1_LDCXXFLAGS = -prof-gen

PASS2_LDCXXFLAGS = $(FASTNOSTATIC) -prof-use

400.perlbench=peak=default:

COPTIMIZE= -ansi-alias

401.bzip2=peak=default:

CC= icc -m64

CPORTABILITY= -DSPEC_CPU_LP64

COPTIMIZE= -opt-prefetch -ansi-alias -auto-ilp32

403.gcc=peak=default:

COPTIMIZE = $(FAST)

feedback=0

429.mcf=peak=default:

COPTIMIZE= $(FAST) -opt-prefetch

feedback=0

#########################################################################################

###########################################

%ifdef %{smt-on}

%ifdef %{physicallogical}

submit = numactl --localalloc --physcpubind=`expr 2 \\* $SPECCOPYNUM` $command

%ifdef %{no-numa)

submit = taskset -c `expr 2 \\* $SPECCOPYNUM` $command

%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}



copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

#########################################################################################

###########################################

445.gobmk=peak=default:

COPTIMIZE= -O2 -ipo -no-prec-div -ansi-alias


PASS2_CFLAGS = $(SSE) -prof-use

PASS1_LDCFLAGS = -prof-gen

PASS2_LDCFLAGS = $(SSE) -prof-use

456.hmmer=peak=default:

CC= icc -m64


COPTIMIZE= $(FAST) -unroll2 -ansi-alias -auto-ilp32

feedback=no

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

#########################################################################################

###########################################

458.sjeng=peak=default:

CC= icc -m64


COPTIMIZE= -unroll4 -auto-ilp32

462.libquantum=peak=default:

CC= icc -m64

CPORTABILITY= -DSPEC_CPU_LP64 -DSPEC_CPU_LINUX



COPTIMIZE= $(FAST) -auto-ilp32 -opt-prefetch

feedback=no

464.h264ref=peak=default:

COPTIMIZE= -unroll2 -ansi-alias

471.omnetpp=peak=default:

CXXOPTIMIZE= -ansi-alias -opt-ra-region-strategy=block

473.astar=peak=default:

CXX= icpc -m64

CXXPORTABILITY= -DSPEC_CPU_LP64

EXTRA_LIBS= -L$(SMARTHEAP64_DIR) -lsmartheap64

CXXOPTIMIZE= -ansi-alias -opt-ra-region-strategy=routine

483.xalancbmk=peak=default:

basepeak=yes

################################################################

# Peak Tuning Flags for FP

################################################################

fp=peak=default:

COPTIMIZE= -auto-ilp32

CXXOPTIMIZE= -auto-ilp32


PASS2_CFLAGS = $(FAST) -prof-use

PASS1_CXXFLAGS = -prof-gen

PASS2_CXXFLAGS = $(FAST) -prof-use

PASS1_FFLAGS = -prof-gen

PASS2_FFLAGS = $(FAST) -prof-use

PASS1_LDFLAGS = -prof-gen

PASS2_LDFLAGS = $(FAST) -prof-use

410.bwaves=peak=default:

OPTIMIZE= $(FAST) -opt-prefetch

feedback=0

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}



copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

%ifdef %{1p-nhm-ex}

copies=8

%endif

%ifdef %{2p-nhm-ex}

copies=16

%endif

%ifdef %{4p-nhm-ex}

copies=32

%endif

#########################################################################################

###########################################

416.gamess=peak=default:

OPTIMIZE= -unroll2 -Ob0 -ansi-alias -scalar-rep-

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

#########################################################################################

###########################################

433.milc=peak=default:

OPTIMIZE= -fno-alias -opt-prefetch

COPTIMIZE=

434.zeusmp=peak=default:

basepeak=yes

435.gromacs=peak=default:

OPTIMIZE= -opt-prefetch



436.cactusADM=peak=default:

basepeak=yes

437.leslie3d=peak=default:

OPTIMIZE= $(FAST)

feedback=no

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

%ifdef %{1p-nhm-ex}

copies=8

%endif

%ifdef %{2p-nhm-ex}

copies=16

%endif

%ifdef %{4p-nhm-ex}

copies=32

%endif

#########################################################################################

###########################################

444.namd=peak=default:

CXXOPTIMIZE= -fno-alias -auto-ilp32

447.dealII=peak=default:

CXXOPTIMIZE= -unroll2 -ansi-alias -scalar-rep-

450.soplex=peak=default:

PORTABILITY =

CXX= icpc -m32

OPTIMIZE= -opt-malloc-options=3

CXXOPTIMIZE=

#########################################################################################

###########################################



%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

#########################################################################################

###########################################

453.povray=peak=default:

CXXOPTIMIZE= -unroll4 -ansi-alias

454.calculix=peak=default:

basepeak=yes

459.GemsFDTD=peak=default:

OPTIMIZE= -unroll2 -Ob0

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6



%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

#########################################################################################

###########################################

465.tonto=peak=default:

OPTIMIZE= -unroll4 -auto -inline-calloc -opt-malloc-options=3

470.lbm=peak=default:

OPTIMIZE= -opt-malloc-options=3 -ansi-alias

#########################################################################################

###########################################

%ifdef %{smt-on}



%ifdef %{no-numa)


%endif

%endif

%endif

%ifdef %{up-dale}

copies=2

%endif

%ifdef %{up-nhm}

copies=4

%endif

%ifdef %{dp-nhm}

copies=8

%endif

%ifdef %{up-wsm-6c}

copies=6

%endif

%ifdef %{dp-wsm-6c}

copies=12

%endif

%ifdef %{1p-nhm-ex}

copies=7

%endif

%ifdef %{2p-nhm-ex}

copies=14

%endif

%ifdef %{4p-nhm-ex}

copies=28

%endif

#########################################################################################

###########################################

481.wrf=peak=default:

basepeak=yes

482.sphinx3=peak=default:

PORTABILITY=

CC= icc -m32

OPTIMIZE= $(FAST)



COPTIMIZE= -unroll2

feedback=no

#################################################################

# (Edit this to match your system)

#################################################################

default=default=default=default:

license_num =

test_sponsor =

hw_avail =

sw_avail =

tester =

hw_cpu_name =

hw_cpu_char =

hw_cpu_mhz =

hw_disk =

hw_fpu =

hw_memory =

hw_model =

hw_ncpuorder =

hw_ncores =

hw_nchips =

hw_ncoresperchip =

hw_nthreadspercore =

hw_other =

hw_pcache =

hw_scache =

hw_tcache =

hw_ocache =

hw_vendor =

prepared_by =

sw_file =

sw_os =

sw_state =

notes_submit_000 = numactl was used to bind copies to the cores

%ifdef %{no-numa)

notes_submit_000 = taskset was used to bind copies to the cores

%endif

int=default=default=default:

sw_compiler000 = Intel C++ Professional Compiler for IA32 and

sw_compiler001 = Intel 64, Version 11.1

sw_compiler002 = Build 20091130 Package ID: l_cproc_p_11.1.064

sw_base_ptrsize = 32-bit

sw_peak_ptrsize = 32/64-bit

sw_other000 = Microquill SmartHeap V8.1

sw_other001 = Binutils 2.18.50.0.7.20080502

fp=default=default=default:

sw_compiler001 = Intel C++ and Fortran Professional Compiler for IA32 and Intel 64,

Version 11.1

sw_compiler002 = Build 20091130 Package ID: l_cproc_p_11.1.064, l_cprof_p_11.1.064

sw_base_ptrsize = 64-bit

sw_peak_ptrsize = 32/64-bit

sw_other001 = Binutils 2.18.50.0.7.20080502



The following section was added automatically, and contains settings that

did not appear in the original configuration file, but were added to the raw file after

the run. default:

flagsurl000 = Intel-ic11.1-linux64-revE.xml



APPENDIX C – SPECint_rate_base2006 OUTPUT FILES




ABOUT PRINCIPLED TECHNOLOGIES

Principled Technologies, Inc. 1007 Slater Road, Suite 250 Durham, NC, 27703 www.principledtechnologies.com

We provide industry-leading fact-based marketing and technology assessment services that help technology vendors and buyers understand the real differences among products. We bring to every assignment extensive experience with and expertise in all aspects of technology marketing, testing, and analysis; from researching new technologies, to developing new methodologies, to testing with existing and new tools, to delivering the results in whatever form best communicates them. When the assessment is complete, we know how to present the results to a broad range of target audiences. We provide our clients with the materials they need, from market-focused data to use in their own collateral to custom sales aids, such as test reports, performance assessments, white papers, PowerPoint presentations, and videos. Every piece of collateral reflects the results of our trusted independent analysis. We customize our services to focus on each client’s requirements. Whether the technology involves hardware, software, Web sites, or services, we offer the experience, expertise, and tools to assess how it will fare against its competition and to highlight its strengths. Our founders, Mark L. Van Name and Bill Catchings, have worked together in technology assessment for 25 years. As journalists, they published over a thousand articles on a wide array of technology subjects. They created and led the Ziff-Davis Benchmark Operation, which developed such industry-standard benchmarks as Ziff Davis Media’s Winstone and WebBench. They founded and led eTesting Labs, and after the acquisition of that company by Lionbridge Technologies were the head and CTO of VeriTest.

Principled Technologies is a registered trademark of Principled Technologies, Inc. All other product names are the trademarks of their respective owners. Disclaimer of Warranties; Limitation of Liability: PRINCIPLED TECHNOLOGIES, INC. HAS MADE REASONABLE EFFORTS TO ENSURE THE ACCURACY AND VALIDITY OF ITS TESTING, HOWEVER, PRINCIPLED TECHNOLOGIES, INC. SPECIFICALLY DISCLAIMS ANY WARRANTY, EXPRESSED OR IMPLIED, RELATING TO THE TEST RESULTS AND ANALYSIS, THEIR ACCURACY, COMPLETENESS OR QUALITY, INCLUDING ANY IMPLIED WARRANTY OF FITNESS FOR ANY PARTICULAR PURPOSE. ALL PERSONS OR ENTITIES RELYING ON THE RESULTS OF ANY TESTING DO SO AT THEIR OWN RISK, AND AGREE THAT PRINCIPLED TECHNOLOGIES, INC., ITS EMPLOYEES AND ITS SUBCONTRACTORS SHALL HAVE NO LIABILITY WHATSOEVER FROM ANY CLAIM OF LOSS OR DAMAGE ON ACCOUNT OF ANY ALLEGED ERROR OR DEFECT IN ANY TESTING PROCEDURE OR RESULT. IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES, INC. BE LIABLE FOR INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES IN CONNECTION WITH ITS TESTING, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES, INC.’S LIABILITY, INCLUDING FOR DIRECT DAMAGES, EXCEED THE AMOUNTS PAID IN CONNECTION WITH PRINCIPLED TECHNOLOGIES, INC.’S TESTING. CUSTOMER’S SOLE AND EXCLUSIVE REMEDIES ARE AS SET FORTH HEREIN.

intel server processors: performance-per-watt · pdf fileintel server processors:...

Documents